Introduction

Cerebral venous thrombosis (CVT) is a rare thrombotic disorder with an incidence of approximately 1.3 per 100,000 per year [1]. Unlike more common locations of venous thromboembolism (VTE), such as deep vein thrombosis (DVT) of the leg and pulmonary embolism (PE), CVT mostly affects young adults and children [2]. Moreover, while the sex ratio is more or less evenly distributed for DVT and PE, CVT predominantly occurs in women, with a sex ratio of 3:1 in adults [3]. Despite these epidemiological differences between CVT and other locations of VTE, there is a substantial overlap in risk factors between both conditions. For instance, cancer, obesity, inflammatory bowel disease, and anemia have all been found to be associated with both CVT and DVT/PE [4]. On the other hand, smoking and immobilization, both established risk factors for DVT/PE, have not been found to increase the risk of CVT.

Factor VIII (FVIII) is a glycoprotein that plays an essential role in the intrinsic pathway of blood coagulation as it functions as a cofactor for activated factor IX [5]. Elevated FVIII has been shown to be an independent risk factor for DVT and PE [6, 7]. The mechanism(s) through which elevated FVIII leads to an increased thrombogenicity are incompletely understood [5]. It is suggested that elevated FVIII by itself is insufficient to cause thrombosis, but that it increases the risk of thrombosis in the presence of other causal factors [6]. Other studies, however, have found that elevated FVIII increases the risk of thrombosis through a direct FVIII-mediated enhancement of thrombin generation [8]. Elevated FVIII has also been found to induce activated protein C resistance [9, 10].

The association between increased FVIII activity and CVT has only been assessed in three controlled studies, but these studies had a very small sample size [11,12,13]. The aim of the current study was to assess whether elevated FVIII activity is a risk factor for CVT.

Methods

Study design and patient selection

We performed a matched case–control study. Cases were adult patients with CVT admitted to the Academic Medical Center in Amsterdam between July 2006 and December 2016. These patients were invited to participate in an observational study on risk factors for CVT. As part of this study, blood was collected during the acute phase (during hospital admission) and in the chronic phase (more than 3 months after diagnosis). For the current study we only used patients for whom we had blood available that was collected in the chronic phase. Controls were healthy hospital staff employees with no history of thrombosis, who donated blood between 2010 and 2013. Cases and controls were matched in a 1:1 ratio for age (within 5 years) and sex.

Diagnosis of CVT was confirmed with CT-venography, MR-venography, catheter angiography, or autopsy, in accordance with international guidelines [14]. For cases, data on demographics, risk factors, clinical manifestations, and outcomes were collected using a standardized case record form. For controls, year of birth, sex and oral contraceptive (OC) use were recorded. Pregnancy at the time of blood collection was not an exclusion criterion. Written informed consent was obtained from all participants or their legal representatives.

Laboratory measurements

Blood samples for measurement of coagulation proteins [including FVIII and von Willebrand factor (vWF) levels] were collected into vacuum tubes containing 0.105 mol/l trisodium citrate and processed within 1 h of collection. Blood was centrifuged at 1500g for 20 min at 15 °C to obtain platelet poor plasma. Then plasma was aliquoted and recentrifuged at 3000g for 15 min at 15 °C, after which it was stored at − 80 °C until the measurements were performed.

Measurement of FVIII activity was performed on an automated coagulation analyzer (Behring Coagulation System) with reagents and protocols from the manufacturer (Siemens Healthcare Diagnostics, Marburg, Germany). vWF antigen levels were determined with an ELISA developed in-house, using antibodies from Dako (Glostrup, Denmark). The reference plasma used in both assays was calibrated to WHO standard 07316 for FVIII activity and vWF antigen. The laboratory measurements were performed simultaneously for cases and controls in one batch at the end of the study.

Statistical analysis

Increased FVIII was defined as a FVIII activity > 150 IU/dl, similar to previous studies [6, 7]. Continuous data were analyzed with a Mann–Whitney test or Student’s t test (whichever was appropriate) and categorical data with a Chi-square test. To determine the association between FVIII and CVT, we used multivariate logistic regression analysis in which we adjusted for age (as a continuous variable) and sex. In a secondary analysis we also adjusted for vWF. We stratified the analyses for sex, and performed several subgroup analyses. The data were analyzed with SPSS version 23.

Results

Case–control study

During the study period, 184 adult patients with CVT were treated at our hospital. We excluded 68 patients because there was no blood collection in the chronic phase (n = 67) or because FVIII measurement failed (n = 1). Therefore, the study population consisted of 116 cases, which were matched with 116 controls. There were no statistically significant differences between included and excluded cases, with the exception of the frequency of cancer and focal neurological deficits (Table 1). Twenty-seven patients (23%) used oral anticoagulation at the time of blood collection. For 30 cases in our study population, blood was available in the acute and chronic phase.

Table 1 Baseline characteristics and risk factors of included and excluded cases
Full size table

Mean age was 40 (SD 11) and 41 (SD 11) years for cases and controls, respectively. In both groups, 99/116 (85%) were women. The median time between CVT diagnosis and blood collection was 18 months [interquartile range (IQR) 7–39 months]. At the time of blood collection, 27 cases (24.8%) used oral anticoagulation.

Median FVIII activity was higher in cases as compared to controls [186 (IQR 159–212) vs. 127 (IQR 102–160), p < 0.001, Fig. 1]. Using the cut-off value of 150 IU/dl, 83.6% of cases had elevated FVIII activity as compared to 28.4% of controls (p < 0.001). After adjustment for sex and age, elevated FVIII substantially increased the risk of CVT (aOR 15.3; 95% CI 7.8–30.1, Table 2). Including vWF as a co-variate only had a minor effect on the strength of the association (aOR 11.6; 95% CI 5.0–27.1). Stratification by sex revealed a stronger association between FVIII and CVT in men (aOR 22.8; 95% CI 2.8–184.3) than in women (aOR 14.7; 95% CI 7.1–30.2, Table 3). Exclusion of cases with a provoked CVT essentially did not change the strength of the association (aOR 16.6; 95% CI 8.0–34.0). The association was stronger in the cases from whom blood was collected within 18 months after CVT (aOR 29.8; 95% CI 10.5–84.4), as compared to those from whom blood was collected after 18 months (aOR 10.1; 95% CI 4.5–22.3). When we excluded patients who were using oral anticoagulation at the time of blood collection, the strength of the association did not substantially change (adjusted OR 12.9; 95% CI 6.3–26.2).

Fig. 1
figure 1

Distribution of FVIII activity in cases and controls. Data are depicted as box and whiskers, with boxes indicating 25th and 75th percentiles and the whiskers the 5th and 95th percentiles. The horizontal line in the box indicates median

Full size image
Table 2 Association between FVIII activity and CVT
Full size table
Table 3 Subgroup analyses
Full size table

Discussion

We found that elevated FVIII is very common in CVT, occurring in more than 80% of patients. In our study, elevated FVIII increased the risk of CVT approximately 15-fold.

Several studies have previously examined FVIII activity in patients with CVT [11,12,13, 15,16,17]. In a cohort study, Aaron et al. found that elevated FVIII was a frequent risk factor in pregnancy associated CVT, occurring in about 15% of patients [15]. Cakmak et al. found a prevalence of 50% for increased FVIII among 16 patients with CVT. In this study, median time between diagnosis and FVIII determination was 8 months [16]. In a study by Bugnicourt et al., elevated FVIII was the most common risk factor for CVT. Their study, which also included 16 patients, demonstrated an elevated FVIII in 25% of patients with CVT. Laboratory screening was carried out a median of 36 months after diagnosis [13]. In an Iranian case–control study which included 25 patients, elevated FVIII increased risk of CVT approximately 10-fold [12]. In another case–control study, Anadure et al. found that elevated FVIII was associated with an 18-fold increase in the risk of CVT [11]. In summary, only three of the previous studies on FVIII and CVT were controlled and all had a very small sample size. Our study is by far the largest study on this topic thus far.

Studies in patients with DVT and PE have found that increased FVIII occurs in about 25% of these patients, and that the presence of elevated FVIII increases the risk of developing these types of VTE approximately fivefold [5,6,7, 18]. Interestingly, elevated FVIII has also been found to be associated with an increased risk of recurrent VTE [7, 19] in a dose-dependent manner [6, 20, 21]. For each 10 IU/dl increment of FVIII activity, the risk for a first or recurrent episode of VTE increases by 10 and 24%, respectively [7]. Based on the data from our study, elevated FVIII appears to occur substantially more often in CVT than other types of VTE. The strength of the association between FVIII and thrombosis also seems to be stronger for CVT than other types of VTE. Our study was underpowered to determine whether elevated FVIII is also associated with recurrent CVT.

The frequency of elevated FVIII among CVT patients in our study is higher than has been reported before, with a prevalence of 83.6%. However, this percentage was found when the cut-off value was > 150 IU/dl. The studies of Bugnicourt, Anadure and Shahsavarzadeh used higher cut-off values, namely > 190, > 170 and > 179 IU/dl respectively [11,12,13]. These differences in used cut-off values of FVIII can explain the differences in calculated percentages and ORs. We had a prevalence of 65.5% of elevated FVIII levels when the cut-off value > 170 IU/dl was used, which is comparable with the prevalence found by Anadure et al. We chose to define increased FVIII as FVIII activity > 150 IU/dl, similar to previous studies [6, 7]. Also, in the Academic Medical Center in Amsterdam in The Netherlands we use > 150 IU/dl in our daily practice as cut-off value.

FVIII is an acute phase protein and is elevated often during the acute stage of a severe illness. For this reason, we only included the cases in whom we could measure FVIII at least 3 months after diagnosis of CVT. Despite this time interval, we still cannot fully exclude the possibility that the elevated FVIII in the cases was a result of the thrombotic event, rather than the cause. Still, given the long interval between diagnosis and blood sampling (median 18 months), this is not a likely scenario. Moreover, previous studies have found that increased FVIII activity persists after VTE, supporting the theory that elevated FVIII levels are independent of the acute phase reaction [7, 20, 22]. Even when we limited our analysis to the cases from whom blood was collected more than 1.5 years after CVT, we still found that elevated FVIII increased the risk of CVT approximately 10-fold.

Administration of adrenaline has been shown to lead to a rapid increase in FVIII activity. This response can be attenuated by beta-blockers [23]. Based on these observations, Hoppener et al. tested the hypothesis whether administration of a beta-receptor blocker lowered elevated FVIII activity in patients with VTE. Indeed, they demonstrated a 23% reduction of FVIII activity 2 weeks after treatment with propranolol and that discontinuation of propranolol resulted in FVIII to return to its initial level [24]. Given the frequency of elevated FVIII in patients with CVT, it would be interesting to determine whether treatment with propranolol also reduces FVIII in patients with CVT and whether this would decrease the risk of recurrent thrombosis.

We found a stronger association between FVIII and CVT in men than in women. The most likely explanation for this observation is the difference in absolute incidence of CVT. In adults, CVT occurs about 3 times more often in women than men [25]. Assuming that elevated FVIII adds a similar absolute risk to the baseline risk of CVT in both men and women, the relative risk would be smaller in women than men [26, 27].

Our study has several limitations. First, approximately one-third of cases were excluded because we had no blood sample available from the chronic phase. While this could introduce a bias, comparison between included and excluded cases did not seem to suggest that this concern is warranted. Second, limited information was available for controls and thus we cannot fully exclude the possibility that the association between FVIII and CVT is partly due to residual confounders. Third, the prevalence of elevated FVIII in the controls was 28%, which is higher than that observed in previous studies [6, 7]. Still, a higher frequency of increased FVIII activity in the control population would only lead to an underestimation of the strength of the association. Another limitation is that our study period of 10 years is long, during which the medical process around CVT has changed. However, there were no changes in the process of collection of blood samples during the study period. Moreover, the laboratory measurements were performed simultaneously for cases and controls in one batch at the end of the study. Therefore we believe the study period has no significant impact on our results. Finally, cases and controls were not recruited in the exact same time period, although the timeframe during which controls were recruited (2010–2013) falls within the time period during which cases were recruited (2006–2016). More importantly, laboratory measurements were carried out simultaneously for cases and controls.

In conclusion, we have found that elevated FVIII is a common and strong risk factor for CVT. Determination of FVIII should be considered in the diagnostic work-up of patients with this condition. Whether elevated factor VIII increased the risk of recurrent thrombosis after CVT remains to be determined.